Extract natural frequencies from acceleration data by FFT

Hi all,

I have problems when calculating frequencies from sample data by FFT in Matlab. The data file shows acceleration in cm/s^2. Sampling frequency is 200 Hz. The file can be downloaded here: http://ifile.it/pqdvhwx. Can you please share a Matlab code for this?

[Arnaud Miege]

Can you show us what you've done so far?

Answer 1

[Bjorn Gustavsson]

At the matlab command line prompt type: help spectrogram

[medo_sk]

I don't think this can help me.

[Bjorn Gustavsson]

Ok, but it would only take you about 30 s to test wouldn't it? I happen to have used that very function to find natural frequencies for sampled signals - as far as I understan this is the first stab-function to use for this purpose - especially if one expects that the dominating frequencies varies.

Answer 2

[medo_sk]

Here is the code done so far. I do not know how to proceed.

% Signal.mat is imported as variable CH1

g=981; %gravity acceleration [cm/s^2]
Fs=200; %Sampling frequency
dt=1/Fs; %Sample time
L=length(CH1); %Length of signal
t=(0:L-1)*dt; %Time vector

CORR=(CH1(:,1)-mean(CH1(:,1)))/g; %Shift of signal to zero value

%Plot Original signal
subplot(311), plot(CH1);
  title('Original Signal');
  xlabel('Samples');
  ylabel('Acceleration [cm/s^2]');

%Plot Original signal in Time Domain
subplot(312), plot(t,CH1);
  title('Original signal in Time Domain');
  xlabel('Time [s]');
  ylabel('Acceleration [cm/s^2]');

%Plot Corrected signal in Time Domain
subplot(313), plot(t,CORR);
  title('Corrected signal in Time Domain');
  xlabel('Time [s]');
  ylabel('Acceleration [cm/s^2]');

n=get(gca,'ytick');
set(gca,'yticklabel',sprintf('%.7f |',n'));

[Arnaud Miege]

OK, so you have pre-processed and plotted your data in the time domain. Now you need to take the FFT of your data using the fft function. Have a look at the documentation, there are a couple of detailed examples. Adapt them to your data:

http://www.mathworks.com/help/releases/R2011a/techdoc/ref/fft.html

[Arnaud Miege]

Also consider formatting your answer so that it's readable:

http://www.mathworks.co.uk/matlabcentral/answers/help/markup

[medo_sk]

Thanks Arnaud for your tips.

If I use the example file and adapt it to my data I do not get reasonable results. I am expecting something different from what is calculated by Y = fft(y,NFFT)/L.

[Arnaud Miege]

Thanks for formatting your answer, much appreciated!! Can you post the code you are using to compute the FFT of your data?